Skip to main content
PMC home page
Journal of Cerebral Blood Flow & Metabolism logoLink to Journal of Cerebral Blood Flow & Metabolism
. 2015 Jan 7;35(4):601–610. doi: 10.1038/jcbfm.2014.233

Neurochemical and BOLD responses during neuronal activation measured in the human visual cortex at 7 Tesla

Petr Bednařík 1,2,3, Ivan Tkáč 1, Federico Giove 4,5, Mauro DiNuzzo 4,5, Dinesh K Deelchand 1, Uzay E Emir 6, Lynn E Eberly 7, Silvia Mangia 1,*
PMCID: PMC4420878  PMID: 25564236

Abstract

Several laboratories have consistently reported small concentration changes in lactate, glutamate, aspartate, and glucose in the human cortex during prolonged stimuli. However, whether such changes correlate with blood oxygenation level–dependent functional magnetic resonance imaging (BOLD-fMRI) signals have not been determined. The present study aimed at characterizing the relationship between metabolite concentrations and BOLD-fMRI signals during a block-designed paradigm of visual stimulation. Functional magnetic resonance spectroscopy (fMRS) and fMRI data were acquired from 12 volunteers. A short echo-time semi-LASER localization sequence optimized for 7 Tesla was used to achieve full signal-intensity MRS data. The group analysis confirmed that during stimulation lactate and glutamate increased by 0.26±0.06 μmol/g (~30%) and 0.28±0.03 μmol/g (~3%), respectively, while aspartate and glucose decreased by 0.20±0.04 μmol/g (~5%) and 0.19±0.03 μmol/g (~16%), respectively. The single-subject analysis revealed that BOLD-fMRI signals were positively correlated with glutamate and lactate concentration changes. The results show a linear relationship between metabolic and BOLD responses in the presence of strong excitatory sensory inputs, and support the notion that increased functional energy demands are sustained by oxidative metabolism. In addition, BOLD signals were inversely correlated with baseline γ-aminobutyric acid concentration. Finally, we discussed the critical importance of taking into account linewidth effects on metabolite quantification in fMRS paradigms.

Keywords: Functional Spectroscopy, GABA quantification, Neurochemistry, Visual Stimulation

Introduction

Functional magnetic resonance spectroscopy (fMRS) is a powerful tool that allows quantifying the dynamic of brain metabolite concentrations in the working brain in vivo. Different laboratories have recently used fMRS at ultra-high magnetic field (7 Tesla (7 T)) to measure the neurochemical responses occurring during stimulation of the human visual cortex.1, 2, 3, 4 The results of these studies were highly consistent with concentration changes in the order of 0.2 μmol/g being reported for aspartate (Asp), glutamate (Glu), glucose (Glc), and lactate (Lac) during prolonged visual stimuli. Similar changes of Glu and Lac have also been reported in the motor cortex.5 The observed functional changes of metabolite concentrations support an overall increase in oxidative energy metabolism during neuronal activation.6 In particular, the opposite changes in Glc and Lac concentrations are thought to reflect increased metabolic rate of Glc utilization and activation of the aerobic glycolytic pathway in brain cells.7, 8, 9 The observed decrease in Asp and increase in Glu have been interpreted as a consequence of an increased rate of the malate-aspartate shuttle, which is associated with the increased flux into the tricarboxylic acid (TCA) cycle.10

The blood oxygenation level–dependent (BOLD) effect measured by functional MRI (BOLD-fMRI) is widely used to map patterns of neuronal activity.11 The BOLD-fMRI signal is linked to neuronal activity through a complex interaction of vascular and metabolic parameters, i.e., cerebral blood flow, cerebral blood volume, and cerebral metabolic rate of oxygen.12 More than two decades since the introduction of the BOLD contrast, the characterization of the neurovascular and metabolic couplings that generate the BOLD-fMRI signal is still at the center of intensive research.13, 14, 15

The present study aimed at characterizing the relationship between BOLD-fMRI signals and the changes in metabolite concentrations during activation of the visual cortex in a group of young, healthy human subjects. BOLD-fMRI signals were also correlated with metabolite concentrations measured at baseline. Our study design benefited from the increased chemical shift dispersion and sensitivity of MRS at 7 T and from using semi-LASER localization sequence to achieve full signal intensity.16 Such an approach allowed robust quantification of metabolite concentrations along with their functional changes not only at a group level as achieved in previous investigations,1, 2, 4 but also on single subjects, thus enabling the quantification of the correlation between fMRI signals and metabolite concentrations in our group of subjects.

The BOLD effect induces a line narrowing of spectra during activation,17 on the order of 0.4 to 0.5 Hz at 7 T.2 Since, LCModel quantification can be slightly biased by such changes in linewidth,18 an artefactual correlation between BOLD signals and metabolite concentration changes is to be expected. In the present study, we therefore assessed the spectral linewidth changes occurring during the functional paradigm for each studied subject rather than at a group level. These subject-specific linewidth changes were taken into account in fMRS data processing to avoid possible biases introduced by the BOLD effect (or other sources of linewidth changes) on metabolite quantification. The BOLD-free concentration changes of metabolites were finally correlated with the amplitude of BOLD-fMRI signals averaged over the same volume of interest (VOI) used for the fMRS measurements in each subject.

Materials and Methods

Participants

Fifteen healthy subjects (7 men, 8 women, and age=33±13 years; mean±s.d.) were enrolled for the study. Exclusion criteria included history of stroke, seizures, neurosurgical procedures, arrhythmias, and severe vision problems. Subjects also met requirements for a study in the magnet, which includes weight <130 kg and the absence of metallic substances in their body. All subjects gave informed consent using procedures, which followed the Code of Federal Regulations, approved by the Institutional Review Board: Human Subjects Committee of the University of Minnesota.

Visual Stimulation

To elicit a strong and robust activation of the visual cortex that can produce measurable changes in metabolite concentrations, a standard block-design functional paradigm with prolonged visual stimulation was used, as in our previous studies.2, 3, 18 During the stimulation period (STIM), the red-black checkerboard, flickering at a frequency of 7.5 Hz and covering the central visual field of view (22 × 29 degrees), was projected on the mirror mounted to the head radio frequency (RF) coil. During the resting condition (REST), only the black background was presented. In addition, a small white cross in the center of the field of view randomly changed its orientation throughout the functional paradigm to keep the subject's attention. Their compliance with the research protocol was monitored during the study by asking them to push a button whenever the cross changed its orientation.

Data Acquisition

MR experiments were performed using a 7 T/90 cm magnet (Agilent/Magnex Scientific, Abington, UK) equipped with a powerful gradient/shim coil (SC72, maximum gradient strength of 70 mT/m; maximum second-order shim strength of 7 mT/m2; Siemens Medical Systems, Erlangen, Germany) that was interfaced to a Siemens Syngo console (Siemens Medical Systems). A half-volume quadrature transceiver RF coil combined with a 8-kW RF amplifier provided the transmit B1+ field just above 40 μT in the center of the occipital lobe.

High-resolution three-dimensional–MPRAGE images (repetition time (TR)=2.5 seconds, echo time (TE)=2.42 ms, inversion time (TI)=1.5 seconds, and isotropic resolution=1 × 1 × 1 mm3) were obtained to visualize the anatomic structure of the occipital cortex. BOLD-fMRI data were acquired with a standard multislice GE-EPI sequence (2.5-mm isotropic resolution, TR=2.5 seconds, TE=23 ms, and 18 slices). The fMRI data (88 time points in total) were collected during the visual stimulation task described above over a total of eight blocks, alternating STIM (7.5 seconds) and REST (20 seconds) conditions, for a total of ~4 minutes of acquisition time. The online general linear model–based statistics was used to enable real-time data analysis.

The functional 1H MRS data were acquired using the full signal-intensity semi-LASER localization sequence16 optimized for 7 T (TE=26 ms and TR=5 seconds), combined with outer volume saturation and VAPOR water suppression.19 The VOI (20 × 20 × 20 mm3) was carefully positioned in the primary visual cortex based on the anatomic landmarks clearly discernible on MPRAGE images and based on the BOLD-fMRI activation maps. The B0-field homogeneities were adjusted by echo-planar version of FASTMAP shimming method.20

To assess the BOLD effect on the water signal, a series of unsuppressed water spectra was acquired during a short visual stimulation session consisting of three blocks REST–STIM–REST. The duration of each block was 30 seconds, and six scans were acquired per block (TR=5 seconds). Finally, the metabolite spectra were collected during a 26.7 minutes fMRS visual stimulation paradigm, consisting of five blocks REST–STIM–REST–STIM–REST. The duration of each block was 5.3 minutes and 64 scans were acquired per block (TR=5 seconds). In addition, unsuppressed water signals were acquired and used as internal reference for absolute metabolite quantification and for correcting the effects of residual eddy currents. The whole acquisition protocol was completed in ~60 minutes.

fMRI Data Analysis

The fMRI data were used not only to select the VOI for fMRS, but also to correlate the amplitude of the BOLD-fMRI signal with the amplitude of the metabolite concentration changes in the same brain volume. Motion correction and statistical analysis of the EPI series were performed with routines integrated in the SPM8 software package (http://www.fil.ion.ucl.ac.uk/spm). No spatial smoothing was applied on the EPI data. General linear model provided statistical parametric maps of the STIM versus REST contrast coefficient and of the constant term (intercept coefficient) in the regression. The VOI position during fMRS was extracted from MRS data files in DICOM format using in-house code written in MATLAB software (http://www.mathworks.com/products/matlab), and a mask was built accordingly. The VOI mask was then used for masking the parametric maps, and for obtaining the mean BOLD-fMRI amplitude, within the VOI used for fMRS. The mean BOLD-fMRI amplitude value was calculated in percentage by dividing, voxel by voxel, the regression coefficient of the STIM condition by the constant term of the regression.

fMRS Data Analysis

Acquired single–scan fMRS data (FIDs) were first corrected for small frequency and phase fluctuations, then summed (over 4 scans or over 32 scans in each subject) and finally corrected for residual eddy currents.21 The intensity of the residual water signal was well below the level of metabolite resonances, hence the water signal removal was not necessary to avoid baseline distortions. The fMRS data summed over four scans per each subject (20 seconds time resolution) were additionally summed across all subjects to increase the signal-to-noise ratio (4 scans × 12 subjects=48 scans per spectra), and thus finally obtain one fMRS data set with high temporal resolution.

Metabolites were quantified from both 32-scan and 48-scan spectra using LCModel (http://www.s-provencher.com/pages/lcmodel.shtml) with the basis set of 19 brain metabolite spectra simulated using a density-matrix approach. A spectrum of fast-relaxing macromolecules was also included in the basis set. This macromolecule spectrum was measured from the occipital lobe of five healthy subjects (not included in the current study) with semi-LASER using metabolite nulling technique (TR=2 seconds and TI=685 ms). The residual signal of phosphocreatine (PCr) at 3.94 p.p.m. with a short T1 was removed by postprocessing and the high-frequency noise was suppressed by the Gaussian filter (σ=0.05 seconds). The unsuppressed water signal was used as an internal reference assuming 80% brain-water content and 9% CSF fraction within the VOI.22 The metabolite concentrations were also corrected for relaxation effects during the echo time of 26 ms using simplified approximations (water T2=64 ms and metabolites T2=107 ms). For all metabolites except Glc, only concentration values quantified with Cramèr-Rao lower bounds (CRLBs) below 50% were included into further analysis. Since the quantification of Glc at 7 T is particularly challenging,23 we used slightly different selection criteria just for this metabolite. Indeed we used the entire Glc time courses of those subjects whose CRLB of Glc were below 50% on an average during the functional protocol, despite the fact that a few data points slightly exceeded this threshold. From all metabolites included in the LCModel basis set, alanine was the only metabolite that was quantified with CRLB <50% only in a small fraction of spectra, and therefore it was excluded entirely from the data analysis.

Finally, similarly to what we have done in our previous visual stimulation studies,2, 3 all fMRS data acquired during the second halves of STIM and after REST periods from all subjects were summed accordingly, resulting in two spectra, STIM and REST, with 768 averages each (thus combined over 32 scans × 2 task blocks × 12 subjects). These two spectra were used to generate the final difference spectrum. To eliminate the BOLD line–narrowing effect, the exponential multiplication (corresponding to 0.46-Hz line broadening) was applied to the STIM FID data set to match the linewidth of the REST spectrum before the calculation of the difference spectrum. Metabolite quantification from the BOLD-free difference spectrum was performed by LCModel analysis with the basis set reduced to four metabolites (Asp, Glc, Glu, and Lac), where two of them (Asp and Glc) were inverted.

Assessment of the BOLD Effect on 1H MR Spectra

The BOLD effect on water signal linewidth was evaluated from MR spectra acquired during the short stimulation paradigm, while the water suppression was turned off. The BOLD line–narrowing effect (ΔFWHMwater) was quantified as a difference between water signal linewidths acquired during STIM and the first REST period, with the difference being calculated as REST−STIM. The first scan of the STIM period was omitted to take into account the delay in the hemodynamic response.

To quantify the linewidth change in metabolite spectra induced by the BOLD effect, the methyl signal of total creatine (Cr) at 3.03 p.p.m. was chosen. The creatine linewidth was determined by decomposing the experimental spectrum near Cr into two components (Cr and a broad signal of macromolecules) using standard spectral fitting built in Varian (Palo Alto, CA, USA) software. Differences in linewidth were quantified between the first STIM and the following REST condition (ΔFWHMCr1) and between the second STIM and the following REST condition (ΔFWHMCr2), with the differences being calculated as REST−STIM. Only spectra from the second halves of STIM and the following REST conditions were used, similarly to what was done for statistical analyses. These values were averaged to obtain the mean Cr-signal linewidth changes for each subject (ΔFWHMCr) during the visual stimulation paradigm. The estimated changes in Cr-signal linewidths were also used to eliminate a possible bias in metabolite quantification caused by the linewidth changes occurring during the long fMRS paradigm. Before the LCModel analysis, the linewidths of spectra acquired during the STIM and subsequent the following REST periods were matched. Specifically, the desired line broadening of STIM spectrum was achieved by exponential multiplication of the FID, corresponding to ΔFWHMCr1 and ΔFWHMCr2 for the first and second pair of STIM and REST periods, respectively. In addition, white noise was added to these line-broadened FIDs to keep the underlying noise level constant.

Statistical Analysis

To determine whether the metabolite changes between STIM and REST conditions were significant, paired two-tailed t-test (α=0.05) was used. To reduce the likelihood of type-1 error (false positives) arising from multiple comparisons (19 metabolites), the false discovery rate method with the q value of 0.05 was applied. To avoid possible transient effects in the metabolic responses, we averaged metabolite levels quantified only from the second half of the two STIM or after REST conditions (32 scans each) during the functional paradigm. Statistical tests were performed on data both with and without the linewidth matching described above.

To assess the statistical relationship between metabolite concentration changes, linewidth changes and BOLD-fMRI amplitudes, Pearson's correlation analysis was applied. In addition, Pearson's correlation analysis with the significance level of 0.05 was used to assess the possible association between baseline concentration of metabolites and the amplitude of the BOLD-fMRI signals induced by the visual stimulation. Only spectra from the second half of the first REST (before stimulation) were used to quantify baseline metabolite concentrations.

Results

Metabolite Concentrations and their Changes during Visual Stimulation

The experimental setup used in this study, specifically the combination of the high magnetic field (7 T) with a highly sensitive RF coil, FASTMAP B0 shimming method and semi-LASER localization sequence, allowed us to acquire fMRS data characterized by high signal-to-noise ratio and high spectral resolution (Figure 1). The BOLD effect on the signals of N-acetylaspartate and Cr+PCr, originating from different linewidths between STIM and REST conditions, is clearly visible on the difference spectrum. The average linewidth of water resonances acquired during the REST condition was 13.6±0.9 Hz (n=11), with concomitant creatine methyl signal linewidths of 11 to 12 Hz. The spectra were artifact free in the entire chemical shift range for the large majority of subjects (Figure 1). The fMRS data of three subjects were discarded because of minor contamination of spectra by unwanted signals from subcutaneous lipids. In all 12 subjects those were included into further analysis, all metabolites except γ-aminobutyric acid (GABA) and Glc were quantified with CRLB <50% from all the 120 spectra (12 subjects, ten 32-scan spectra per subject) acquired throughout the visual stimulation paradigm. The CRLBs of GABA quantification exceeded 50% only in 3 of 120 cases. Using slightly modified selection rules for Glc (see Materials and Methods section), the average CRLB <50% criteria was fulfilled in 8 of 12 subjects. The entire Glc time courses of these eight subjects were included into further analysis. The CRLBs of Glc exceeded 50% only in 2 of 80 cases (eight subjects, 10 spectra per subject), and were below 30% on an average. Overall, average CRLB were below 10% for Asp, Cr, PCr, glutathione, glutamine, Glu, myo-inositol, N-acetylaspartate, N-acetylaspartylglutamate, glycerophosphocholine, phosphoethanolamine, phosphocholine, and Lac. The remaining weakly represented metabolites, such as ascorbate, GABA, scyllo-inositol, and taurine had CRLBs below 20%. The values of CRLBs in concentration units were <0.2 μmol/g for most of the metabolites and below 0.1 μmol/g for Lac. The baseline resting–state metabolite concentrations with associated CRLBs are summarized in Table 1.

Figure 1.

Figure 1

Open in a new tab

Representative 1H MR spectra acquired from the occipital cortex during visual stimulation (A) and rest (B) periods. The typical pattern (narrow linewidth, sideband undershoot) of small residual signals of NAA (2 p.p.m.) and total creatine (3 p.p.m.) in the difference spectrum (C) clearly indicates the BOLD effect, which reduces the linewidth of spectra acquired during the stimulation period. MRS parameters: semi-LASER, TE=26 ms, TR=5 seconds, number of scans=32, data acquired during second halves of STIM, and REST periods. Inset: fMRI—typical VOI selection in the activated brain region. BOLD, blood oxygenation level dependent; Cr, creatine; Gln, glutamine; Glu, glutamate; GPC, glycerophosphocholine; Ins, inositol; Lac, lactate; NAA, N-acetylaspartate; PC, phosphocholine; PCr, phosphocreatine; REST, resting condition; STIM, stimulation period; VOI, volume of interest.

Table 1. Group analysis of metabolite concentrations and their changes during the visual stimulation paradigm.

  Baseline concentration CRLB Concentration difference (STIM−REST)
  Mean±s.e.m. (μmol/g) Mean (%) No linewidth correction
 After linewidth matching
      Mean (%) Mean±s.e.m. (μmol/g) Mean (%) Mean±s.e.m. (μmol/g)
Asc 0.96±0.04 16.9 1.7 0.01±0.03 −4.3 −0.05±0.03
Asp 3.58±0.15 7.4 −4.5 −0.17±0.04** −5.4 −0.20±0.04***
Cr 4.22±0.10 3.7 2.2 0.09±0.03* 1.0 0.04±0.02
Cr+PCr 7.57±0.14 1.0 1.7 0.13±0.03* 0.9 0.06±0.02
GABA 1.03±0.09 16.6 0.9 −0.01±0.03 3.7 0.02±0.03
Glc 0.62±0.14 26.8 −16.7 −0.19±0.03** −16.0 −0.19±0.03**
Gln 2.79±0.12 5.3 0.2 0.00±0.02 1.2 0.03±0.02
Glu 8.59±0.15 2.0 4.0 0.34±0.04*** 3.3 0.28±0.03***
GPC 0.54±0.02 9.0 1.6 0.01±0.01 1.7 0.01±0.01
GSH 1.09±0.03 7.0 2.9 0.03±0.02 2.8 0.03±0.02
Lac 1.01±0.07 9.1 30.0 0.26±0.06** 29.6 0.26±0.06**
myo-Ins 6.08±0.18 2.0 2.1 0.13±0.04** 1.1 0.07±0.04
NAA 11.90±0.20 1.0 0.5 0.06±0.03 0.0 0.00±0.03
NAAG 1.32±0.07 6.5 −1.0 −0.01±0.01 −2.6 −0.03±0.01*
PC 0.40±0.01 11.8 2.5 0.01±0.01 0.7 −0.00±0.01
PCr 3.34±0.07 4.6 1.2 0.04±0.03 0.9 0.03±0.03
PE 0.93±0.03 9.2 4.6 0.04±0.01* 4.9 0.04±0.01*
scyllo-Ins 0.27±0.04 11.9 3.0 0.01±0.01 3.0 0.01±0.01
Tau 1.27±0.08 10.5 5.2 0.05±0.03 2.9 0.02±0.03
Open in a new tab

Abbreviations: Asc, ascorbate; Asp, aspartate; Cr, creatine; CRLB, Cramèr-Rao lower bound; fMRS, functional magnetic resonance spectroscopy; GABA, γ-aminobutyric acid; Glc, glucose; Gln, glutamine; Glu, glutamate; GPC, glycerophosphocholine; GSH, glutathione; Lac, lactate; myo-Ins, myo-inositol; NAA, N-acetylaspartate; NAAG, N-acetylaspartylglutamate; PC, phosphocholine; PCr, phosphocreatine; PE, phosphoethanolamine; REST, resting condition; scyllo-Ins, scyllo-inositol; STIM, stimulation period; Tau, taurine. Metabolite baseline concentrations and related CRLBs were quantified from fMRS data acquired during the second half of the first REST period (before stimulation). Mean concentration differences (STIM−REST) and their statistical significance were calculated from the second halves of STIM and after REST periods. Concentration changes were assessed in two different ways: (a) directly from fMRS data acquired during STIM and REST having slightly different linewidth (central column) and (b) from the fMRS data after linewidth matching between STIM and REST conditions (right column). Statistical significance of the concentration differences (STIM−REST) was assessed with two-tailed paired t-test combined with the false discovery method (q=0.05) to reduce the probability of false-positive results (*P<0.02, **P<0.002, ***P<0.

0005; N=8 for Glc and N=12 for all other metabolites).

During the functional paradigm, the temporal changes in Glu and Lac concentrations (32-scan blocks, 2.7 minutes time resolution) were discernible in individual subjects (Figure 2A). The high temporal resolution (20 seconds) of Glu and Lac changes was achieved by analyzing fMRS data summed over all subjects (4 scans × 12 subjects=48 scans per time point; Figure 2B). The time courses of group-average concentration changes in Glu, Asp, Lac, and Glc (32 scans, 2.7 minutes resolution) are shown in Figure 3. The concentration differences in relative (%) as well as in absolute units (μmol/g) between STIM and REST conditions and their significance are presented in Table 1. Significant differences between STIM and REST conditions were found for eight metabolites (Asp, Cr, Cr+PCr, Glc, Glu, Lac, myo-inositol, and Lac) when using the MRS data uncorrected for the BOLD line–narrowing effect. The identical statistical assessment was performed on MRS data corrected for the BOLD line–narrowing effect by matching the spectral linewidths of all STIM and subsequent REST periods pairs (ΔFWHMCr1, ΔFWHMCr2). This linewidth-matching approach changed the differences between STIM and REST conditions only very modestly, by less than ±0.07 μmol/g for all quantified metabolites. However, these corrections had major consequences on the results of statistical evaluation as the number of metabolites with significant changes dropped to five (Table 1). The changes in Asp, Glc, Glu, and Lac concentrations in the range of ±0.2 to 0.3 μmol/g remained highly significant. The corrected change in phosphoethanolamine was 0.04 μmol/g, almost one order of magnitude smaller than changes observed for Asp, Glc, Glu, and Lac, and barely stayed significant at false discovery rate; q value set to 0.05.

Figure 2.

Figure 2

Open in a new tab

(A) Time courses of glutamate and lactate concentrations during the visual stimulation paradigm (number of scans=32, 2.7 minutes resolution) in individual subjects (N=12). The MRS data acquired in the second halves of the STIM and REST periods (black points) were used to calculate concentration differences between the STIM and REST conditions and for statistical analysis of fMRS data. (B) Time courses of glutamate and lactate concentrations with high temporal resolution. The fMRS data of four scans (20 seconds) were summed across all subjects (N=12). The resulting spectra (4 scans × 12 subjects=48 scans per time point) were quantified by LCModel. The error bars indicate CRLBs. CRLBs, Cramèr-Rao lower bounds; fMRS, functional magnetic resonance spectroscopy; REST, resting condition; STIM, stimulation period.

Figure 3.

Figure 3

Open in a new tab

Time courses of glutamate (Glu), aspartate (Asp), lactate (Lac), and glucose (Glc) concentrations during the visual stimulation paradigm averaged across subjects (N=12 for Glu, Asp, and Lac; N=8 for Glc). Metabolite concentrations were quantified from the 32 scansums (2.7 minutes time resolution). Error bars indicate s.e.m., while shaded areas indicate the stimulation (STIM) periods. P-values correspond to statistical evaluation of differences between STIM and subsequent REST (resting) periods (paired t-tests, mean values from second half of each period).

Similarly to what we have done in our previous fMRS study,2 the linewidth-matched difference spectrum between STIM and REST conditions generated from all subjects was analyzed by LCModel (Figure 4). The obtained concentration changes of Asp, Glc, Glu, and Lac were in excellent agreement with the results of the group-average differences between STIM and REST conditions observed on the individual metabolite time courses (Figure 3; Table 1).

Figure 4.

Figure 4

Open in a new tab

LCModel analysis of the difference spectrum. (A) Magnetic resonance spectroscopy data acquired during the second half of either STIM or REST periods of all subjects were summed accordingly (32 scans × 2 blocks × 12 subjects=768 scans in total per condition) and then subtracted. (B) Difference spectrum without the BOLD effect: the spectral linewidth of summed STIM and REST data was matched (line broadening of the STIM spectrum by 0.46 Hz) before subtraction. (C) LCModel fit of the difference spectrum without the BOLD effect. (D) Residual of the LCModel analysis. (E) 1H MR spectra of individual metabolites provided by LCModel analysis and quantified metabolite concentrations of the difference spectrum (mean±CRLB). Gaussian filter (σ=0.1 seconds) was applied on the spectra (A, B, and D) only for display purposes. Asp, aspartate; BOLD, blood oxygenation level dependent; CRLB, Cramèr-Rao lower bound; Glc, glucose; Glu, glutamate; Lac, lactate; REST, resting condition; STIM, stimulation period.

Correlations between BOLD Response and Changes in Metabolite Concentrations

The mean amplitude of the BOLD-fMRI signal measured in the fMRS VOI was (2.4%±1.2% n=11, mean±s.d.) resulting in intersubject variability of ~48% (s.d./mean). The volume of activated voxels in the whole fMRI field of view (at threshold T=3.19, P<0.001) was characterized by a similar intersubject variability (27±11 mL activated voxels, coefficient of variation=41%). The mean fraction of activated voxels within the VOI used for fMRS was quite consistent among subjects, namely 45.2%±18.5%. Altogether, these observations suggest that the intersubject variability of the BOLD-fMRI signal is likely not because of inconsistent fMRS VOI positioning, but rather because of differences in responses among subjects.

Changes in water signal linewidth (ΔFWHMwater) determined from the short stimulation paradigm (1.5 minutes) were compared with the changes in total Cr linewidth (ΔFWHMCr) determined from the full functional paradigm (26.7 minutes). In 6 of 11 subjects, ΔFWHMwater and ΔFWHMCr were spread close to the line of identity (Figure 5A). In the remaining five subjects, changes in Cr linewidth were slightly smaller (<0.4 Hz) than ΔFWHMwater, indicating that factors other than the BOLD effect altered the spectral linewidth throughout the entire fMRS experiment, such as B0-shim stability and minor subject movements. As expected, the water signal linewidth changes acquired during the short stimulation paradigm strongly correlated with the overall BOLD-fMRI amplitude in the selected VOI (N=10, R=0.93, P=0.0001; Figure 5B).

Figure 5.

Figure 5

Open in a new tab

Relationships between spectral linewidth changes, blood oxygenation level–dependent functional magnetic resonance imaging (BOLD-fMRI) amplitudes, metabolite concentrations, and their changes during the visual stimulation studies. (A) Scatter plot showing linewidth changes of creatine (Cr) signals (3 p.p.m.) between REST (resting) and STIM (stimulation) conditions (ΔFWHMCr) assessed from the 27-minute functional magnetic resonance spectroscopy (fMRS) paradigm, relative to linewidth changes of water signals between REST and STIM conditions (ΔFWHMwater) determined from the short stimulation paradigm (1.5 minutes). Dashed line represents the line of identity. (B) Correlation between water signal–linewidth changes (REST−STIM) and BOLD signal amplitudes (averaged across the volume selected for fMRS) assessed from the fMRI data (R=0.93, P=0.0001). (C and D) Correlation of BOLD-fMRI amplitudes with the relative concentration changes of glutamate (Glu; R=0.73, P=0.01; C) and lactate (Lac; R=0.65, P=0.003; D) in response to visual stimulation. (E) Correlation between the baseline resting–state γ-aminobutyric acid (GABA) concentration and the BOLD-fMRI amplitudes (R=−0.65, P=0.043).

Notably, the BOLD-fMRI amplitudes were significantly and positively correlated with the relative concentration changes of Glu (N=11, R=0.73, P=0.01) and Lac (N=11, R=0.65, P=0.003) quantified using the STIM–REST linewidth-matching approach (Figures 5C and 5D). No significant correlations were observed between BOLD-fMRI signals and concentration changes of any other metabolites. The correlation between the BOLD-fMRI amplitudes and uncorrected concentration differences of Glu were apparently stronger (R=0.87, P=0.0005), because of the slight bias of LCModel quantification caused by changes in linewidth. When absolute changes of metabolite concentrations were considered, the correlation between the BOLD-fMRI amplitudes and Δ[Glu] was also significant (R=0.67, P=0.024), whereas only a trend for a weak correlation was observed for Δ[Lac] (R=0.48, P=0.13). The linear regression of Δ[Glu] versus BOLD-fMRI data (Figure 5C) revealed a trend toward a nonzero intercept value Δ[Glu]0=1.4%±0.7% (i.e., 0.14±0.06 μmol/g), which almost reached statistical significance (P=0.06).

Correlations between BOLD Response and Resting State Metabolite Levels

The correlation analysis was performed also between the BOLD-fMRI amplitudes and baseline metabolite concentrations to investigate a possible relationship between these factors. A significant inverse correlation between the BOLD-fMRI amplitude and the resting-state concentration of GABA was found (R=−0.65, P=0.043, N=10; Figure 5E). Significant correlations of the BOLD amplitude with the baseline concentration of other metabolites were not observed.

Discussion

Blood oxygenation level–dependent signals and metabolite concentration changes are distinct manifestations of increased neuronal activity. The BOLD signals are inherently defined relative to a baseline (in % units), and they rely on a complex interplay of hemodynamic (cerebral blood flow and cerebral blood volume) and metabolic (cerebral metabolic rate of oxygen) parameters. However, metabolite concentration changes can be quantified in absolute units, and they result from altered production/consumption balances because of changes in metabolic rates (e.g., glycolysis rate, TCA cycle rate, and cerebral metabolic rate of oxygen). Even if there is no intuitive causal link between BOLD signals and metabolite concentration changes, a correlation between them can be anticipated because they both are strongly related to neuronal activity and they both are affected by changes in oxidative metabolism. However, whether and how much metabolite concentrations changes are actually correlated with BOLD signals is nontrivial to predict. In fact, BOLD signals only partly reflect changes in metabolism, as the BOLD effect occurs purely in the presence of a mismatch between changes in metabolism (cerebral metabolic rate of oxygen) and hemodynamic (cerebral blood flow and cerebral blood volume). However, changes in metabolite steady-state concentrations are not biased by hemodynamics, and may thus reflect changes in metabolism more directly than BOLD. The primary aim of this study was to establish whether metabolite concentration changes correlate with BOLD signals during increased neuronal activity. To answer this question, we adopted a similar study design and experimental setup to that used in our previous investigations, except for the use of a full signal-intensity semi-LASER localization sequence that provided an almost twofold increase in signal-to-noise ratio in comparison to STEAM localization sequence.2 This choice allowed a reliable detection of Glu and Lac time courses in single subjects. To minimize possible bias caused by the BOLD line–narrowing effect on metabolite quantification,2, 17 we also quantified functional changes in metabolite levels using linewidth-matched STIM and REST spectroscopy data. The group analysis results (Table 1; Figures 3 and 4) were in excellent agreement with previous studies from our group (conducted with a different localization sequence)2 as well as from other laboratories.1, 4 At a single-subject level, positive correlations of Glu and Lac concentration changes with BOLD signals were observed for the first time. The BOLD effect is a reliable surrogate of increased neuronal activation in the visual cortex, especially during paradigms that unambiguously increase neuronal excitation like the one used in this study. Therefore, our findings of positive correlations between BOLD signals and Glu and Lac changes are consistent with Glu and Lac reflecting increased energy demands during higher neuronal activation of the visual cortex. Finally, we observed an inverse correlation between baseline GABA and BOLD signals, in agreement with previous findings obtained with edited single-voxel spectroscopy.24, 25

The BOLD effect is associated with a line narrowing of the water signal.17 The narrowing effect can also be observed on the singlets of MR spectra like those of Cr and N-acetylaspartate, and can introduce a small bias in LCModel quantification.18 Even though these effects are minor, they cannot be neglected if one intends to determine correlation between changes in metabolite concentrations and the BOLD effect. In our study, we estimated the linewidth changes induced by the BOLD effect by using two independent approaches. In the first one, we assessed linewidth changes of the water signal acquired during a short stimulation paradigm. In this case, changes in water linewidth were related mainly to the BOLD effect, as confirmed by their excellent correlation with fMRI signals (Figure 5B). In the second approach, we assessed the linewidth changes of the singlet of Cr at 3 p.p.m. during the 26.7-minute fMRS experiment. When plotting the linewidth changes of water and Cr peaks measured in each subject during a short and long fMRS paradigm, respectively, they fall on the identity line in most but not all subjects (Figure 5A). Such a finding implies that for most subjects the BOLD effect was the main cause of linewidth changes also during the long fMRS paradigm. However, in some subjects, the changes in linewidth originated from factors other than the BOLD effect, such as subtle motions of the subject or long-term B0-field/-shim instabilities.

To minimize possible biases of LCModel quantification induced by linewidth changes in each subject, we opted to take into account linewidth changes estimated from the Cr peak during the long fMRS paradigm itself. Overall, the linewidth correction affected LCModel quantification by only a few percent (Table 1). However, whereas before linewidth corrections eight metabolites showed statistically significant changes in concentration during the fMRS paradigm, after linewidth correction the changes in Cr, Cr+PCr, and myo-inositol lost statistical significance, and phosphoethanolamine barely remained significant with a concentration change much smaller than what was observed for Glu, Asp, Glc, and Lac. Notably, the statistical significance and extent of concentration changes of Glu, Asp, Glc, and Lac were only partially affected by the linewidth correction, confirming that the results for those four metabolites reflect real concentration changes. No robust and significant changes were found for any other metabolite, despite using a more sensitive MRS sequence than what we used in the past. Whereas different groups have largely confirmed changes of Glu, Asp, Glc, and Lac, some authors have additionally reported small variations in concentration of other metabolites during the activation of the visual cortex. In particular, Lin et al1 reported an increase of glutathione and glutamine, and a decrease in glycine (Gly), whereas Baslow et al26 reported a decrease of N-acetylaspartate during long visual stimuli. Differences in stimulus characteristics (e.g., chromatic versus achromatic stimuli) or acquisition schemes (e.g., short versus long echo time) might potentially contribute to slightly discrepant results on metabolites other than Glu, Asp, Glc, and Lac.

In the present study, the amplitude of Glu and Lac concentration changes was found to be positively correlated with BOLD-fMRI signals (Figures 5C and 5D). However, correlation of BOLD signals with Glc and Asp changes was not significant, likely because of the small sample size for Glc (N=7), and to CRLBs comparable with the size of the group effect for Asp (7.4% versus 5.5% Table 1), which limited the sensitivity of the measurement at a single-subject level. The correlation of BOLD signals with Glu and Lac concentration changes can be interpreted in the context of increased oxidative energy metabolism as occurring in response to the increased energy demands of neuronal activation. We have previously suggested Lac to reflect aerobic glycolysis through the exchange of Lac with the pyruvate pool.2, 3 Here we suggest that Glu plays a similar role, through the exchange of Glu with the α-ketoglutarate pool (Figure 6). The observed changes in Asp are compatible with this scenario, since Asp is directly involved in the transamination reaction converting α-ketoglutarate into Glu (Figure 6). The notion that the increase in Glu during sensory stimuli is a sign of increased TCA cycle rate is overall consistent with several lines of evidence, as we previously described. 27 Importantly, ΔGlu might be generally easier to measure than ΔLac.

Figure 6.

Figure 6

Open in a new tab

Suggested scheme of changes in metabolic fluxes and steady-state concentration of brain metabolites during increased brain activity induced by visual stimulation. Increased brain activity requires an increased energy production rate, which is achieved by an increase in glycolytic flux (accompanied by increased cerebral metabolic rate of glucose, CMRGlc) and by an increase in tricarboxylic acid (TCA) cycle and mitochondrial respiration flux (accompanied by increased cerebral metabolic rate of oxygen (CMRO2)). Arrows indicate changes in the steady-state concentration of intermediate brain metabolites and metabolites detectable by magnetic resonance spectroscopy (bold fonts). The decrease in glucose (Glc) is caused by increased CMRGlc at a limited transport rate from plasma, changes in lactate (Lac) and glutamate (Glu) concentrations follow changes in steady-state concentration of pyruvate and α-ketoglutarate (α-KG), respectively, because of their dynamic exchange. Aspartate (Asp) decrease is associated with increased transamination reaction to Glu.

The results of the present study show a trend for a nonzero intercept of ΔGlu when plotted against BOLD signals. This is an interesting observation, which warrants future studies with larger cohorts of subjects. Indeed, if confirmed, this finding would suggest different sensitivities of fMRI and fMRS signals to the underlying vascular and metabolic events when evaluated in large cortex volumes. Interestingly, dissociation between the BOLD effect and the hemodynamic response (cerebral blood flow) has been observed under hypercapnia.28 In principle, it is generally conceivable that an activated brain area shows no BOLD effect if changes in blood flow do not overshoot changes in oxygen consumptions, but shows increased Glu levels because of the increased energy demands of neuronal activation. A trend for a nonzero intercept was not observed in the linear regression analysis between ΔLac and BOLD signals (Figure 5D). This subtle difference between ΔGlu and ΔLac could simply result from a slightly decreased detection sensitivity of measuring Lac relative to Glu (Figure 2). But if confirmed by further studies, such finding would suggest that Glu and Lac may capture increased oxidative energy metabolism in different ways, possibly because of the distinct stages of the oxidative pathways that Glu and Lac are linked to. A great deal of caution is nevertheless required when interpreting the relationships of ΔGlu and ΔLac with BOLD signals. In fact, ΔGlu and ΔLac as well as BOLD signals are likely strongly dependent on network activity (e.g., balance between excitation and inhibition) brought about by specific types of sensory stimulation.

The large interindividual variability of the stimulus-evoked fMRI signal observed in our study (Figure 5B) can be caused by factors that are of nonneural origin,29, 30, 31 and by variations in the baseline brain activity.32, 33 Interestingly, the inverse correlation between BOLD signals and baseline GABA concentrations observed here appears to substantiate the role of inhibitory/excitatory balance in modulation of BOLD-fMRI responses, as also suggested by similar findings obtained at 3 T.24, 25, 34 Nonetheless, the high degree of homeostasis between GABAergic and glutamatergic activity advances the possibility that the inverse correlation of baseline GABA and BOLD signals does not concern the commonly accepted notion that increase (or decrease) in inhibition translates to decrease (or increase) in excitation.35 Rationalizing these conflicting findings, and in general the role of subject-specific baseline variables in evoked brain activity, will require further research (for details, see Shulman et al).36

Finally, it is worth noting that concentration changes measured from large regions of interest as performed in the present study should be evaluated with caution, and partial volume effects should be taken into account. In our subjects, 45.2%±18.5% of voxels within the spectroscopy VOI reached the activation threshold for BOLD signals. Such an observation, albeit threshold dependent, suggests that functional concentration differences might be underestimated because nonactivated voxels are included in the VOI.

Conclusions

We conclude that variations in concentrations of Glu, Asp, Glc, and Lac during prolonged visual stimuli are robust markers of neuronal activation in the human visual cortex. Changes in Glu and Lac concentrations positively correlate with BOLD signals, and both reflect increased energy demands fulfilled by oxidative metabolism. Baseline GABA concentration is inversely correlated with BOLD signals, and possibly contributes to the large variability of BOLD responses observed among healthy subjects. In addition, during long fMRS paradigms, linewidth changes of metabolite peaks occur as a result of the BOLD effect and possibly of other factors not related to BOLD. Such linewidth changes can introduce minor bias in LCModel quantification, and they should be taken into account when reliable quantification of minute concentration changes is desired.

Acknowledgments

The authors thank their research volunteers for their participation, and to Michelle Hartwig for helping with subject recruitment. The authors also thank Andrea Grant for helping with the visual stimulation setup, and Dr Cheryl Olman for valuable discussions.

The authors declare no conflict of interest.

Footnotes

This work was supported by the CEITEC (CZ.1.05/1.1.00/02.0068) from the European Regional Development Fund, and by the grants NIH 1R03NS082541 (to SM), NIH P41 RR008079, P41 EB015894 and P30 NS076408 (to Center for Magnetic Resonance Research). Research reported in this publication was also supported by the National Center for Advancing Translational Sciences of the National Institutes of Health Award Number UL1TR000114. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

References

  1. Lin Y, Stephenson MC, Xin L, Napolitano A, Morris PG. Investigating the metabolic changes due to visual stimulation using functional proton magnetic resonance spectroscopy at 7 T. J Cereb Blood Flow Metab. 2012;32:1484–1495. doi: 10.1038/jcbfm.2012.33. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Mangia S, Tkac I, Gruetter R, Van de Moortele PF, Maraviglia B, Ugurbil K. Sustained neuronal activation raises oxidative metabolism to a new steady-state level: evidence from 1H NMR spectroscopy in the human visual cortex. J Cereb Blood Flow Metab. 2007;27:1055–1063. doi: 10.1038/sj.jcbfm.9600401. [DOI] [PubMed] [Google Scholar]
  3. Mangia S, Tkac I, Logothetis NK, Gruetter R, Van de Moortele PF, Ugurbil K. Dynamics of lactate concentration and blood oxygen level-dependent effect in the human visual cortex during repeated identical stimuli. J Neurosci Res. 2007;85:3340–3346. doi: 10.1002/jnr.21371. [DOI] [PubMed] [Google Scholar]
  4. Schaller B, Mekle R, Xin L, Kunz N, Gruetter R. Net increase of lactate and glutamate concentration in activated human visual cortex detected with magnetic resonance spectroscopy at 7 Tesla. J Neurosci Res. 2013;91:1076–1083. doi: 10.1002/jnr.23194. [DOI] [PubMed] [Google Scholar]
  5. Schaller B, Xin L, O'Brien K, Magill AW, Gruetter R. Are glutamate and lactate increases ubiquitous to physiological activation? A (1)H functional MR spectroscopy study during motor activation in human brain at 7 Tesla. Neuroimage. 2014;93 Pt 1:138–145. doi: 10.1016/j.neuroimage.2014.02.016. [DOI] [PubMed] [Google Scholar]
  6. Mangia S, Giove F, Tkac I, Logothetis NK, Henry PG, Olman CA, et al. Metabolic and hemodynamic events after changes in neuronal activity: current hypotheses, theoretical predictions and in vivo NMR experimental findings. J Cereb Blood Flow Metab. 2009;29:441–463. doi: 10.1038/jcbfm.2008.134. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. DiNuzzo M, Mangia S, Maraviglia B, Giove F. Changes in glucose uptake rather than lactate shuttle take center stage in subserving neuroenergetics: evidence from mathematical modeling. J Cereb Blood Flow Metab. 2010;30:586–602. doi: 10.1038/jcbfm.2009.232. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Gjedde A, Marrett S. Glycolysis in neurons, not astrocytes, delays oxidative metabolism of human visual cortex during sustained checkerboard stimulation in vivo. J Cereb Blood Flow Metab. 2001;21:1384–1392. doi: 10.1097/00004647-200112000-00002. [DOI] [PubMed] [Google Scholar]
  9. Simpson IA, Carruthers A, Vannucci SJ. Supply and demand in cerebral energy metabolism: the role of nutrient transporters. J Cereb Blood Flow Metab. 2007;27:1766–1791. doi: 10.1038/sj.jcbfm.9600521. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. McKenna MC, Waagepetersen HS, Schousboe A, Sonnewald U. Neuronal and astrocytic shuttle mechanisms for cytosolic-mitochondrial transfer of reducing equivalents: current evidence and pharmacological tools. Biochem Pharmacol. 2006;71:399–407. doi: 10.1016/j.bcp.2005.10.011. [DOI] [PubMed] [Google Scholar]
  11. Ogawa S, Tank DW, Menon R, Ellermann JM, Kim SG, Merkle H, et al. Intrinsic signal changes accompanying sensory stimulation: functional brain mapping with magnetic resonance imaging. Proc Natl Acad Sci USA. 1992;89:5951–5955. doi: 10.1073/pnas.89.13.5951. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Giove F, Mangia S, Bianciardi M, Garreffa G, Di Salle F, Morrone R, et al. The physiology and metabolism of neuronal activation: in vivo studies by NMR and other methods. Magn Reson Imaging. 2003;21:1283–1293. doi: 10.1016/j.mri.2003.08.028. [DOI] [PubMed] [Google Scholar]
  13. Buxton RB. Interpreting oxygenation-based neuroimaging signals: the importance and the challenge of understanding brain oxygen metabolism. Front Neuroenergetics. 2010;2:8. doi: 10.3389/fnene.2010.00008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Buxton RB, Griffeth VE, Simon AB, Moradi F. Variability of the coupling of blood flow and oxygen metabolism responses in the brain: a problem for interpreting BOLD studies but potentially a new window on the underlying neural activity. Front Neurosci. 2014;8:139. doi: 10.3389/fnins.2014.00139. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Kim SG, Ogawa S. Biophysical and physiological origins of blood oxygenation level-dependent fMRI signals. J Cereb Blood Flow Metab. 2012;32:1188–1206. doi: 10.1038/jcbfm.2012.23. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Oz G, Tkac I. Short-echo, single-shot, full-intensity proton magnetic resonance spectroscopy for neurochemical profiling at 4 T: validation in the cerebellum and brainstem. Magn Reson Med. 2011;65:901–910. doi: 10.1002/mrm.22708. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Zhu XH, Chen W. Observed BOLD effects on cerebral metabolite resonances in human visual cortex during visual stimulation: a functional (1)H MRS study at 4 T. Magn Reson Med. 2001;46:841–847. doi: 10.1002/mrm.1267. [DOI] [PubMed] [Google Scholar]
  18. Mangia S, Tkac I, Gruetter R, Van De Moortele PF, Giove F, Maraviglia B, et al. Sensitivity of single-voxel 1H-MRS in investigating the metabolism of the activated human visual cortex at 7 T. Magn Reson Imaging. 2006;24:343–348. doi: 10.1016/j.mri.2005.12.023. [DOI] [PubMed] [Google Scholar]
  19. Tkac I, Starcuk Z, Choi IY, Gruetter R. In vivo 1H NMR spectroscopy of rat brain at 1 ms echo time. Magn Reson Med. 1999;41:649–656. doi: 10.1002/(sici)1522-2594(199904)41:4<649::aid-mrm2>3.0.co;2-g. [DOI] [PubMed] [Google Scholar]
  20. Gruetter R, Tkac I. Field mapping without reference scan using asymmetric echo-planar techniques. Magn Reson Med. 2000;43:319–323. doi: 10.1002/(sici)1522-2594(200002)43:2<319::aid-mrm22>3.0.co;2-1. [DOI] [PubMed] [Google Scholar]
  21. Klose U. In vivo proton spectroscopy in presence of eddy currents. Magn Reson Med. 1990;14:26–30. doi: 10.1002/mrm.1910140104. [DOI] [PubMed] [Google Scholar]
  22. Emir UE, Raatz S, McPherson S, Hodges JS, Torkelson C, Tawfik P, et al. Noninvasive quantification of ascorbate and glutathione concentration in the elderly human brain. NMR Biomed. 2011;24:888–894. doi: 10.1002/nbm.1646. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Tkac I, Oz G, Adriany G, Ugurbil K, Gruetter R. In vivo 1H NMR spectroscopy of the human brain at high magnetic fields: metabolite quantification at 4T vs. 7T. Magn Reson Med. 2009;62:868–879. doi: 10.1002/mrm.22086. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Muthukumaraswamy SD, Edden RA, Jones DK, Swettenham JB, Singh KD. Resting GABA concentration predicts peak gamma frequency and fMRI amplitude in response to visual stimulation in humans. Proc Natl Acad Sci USA. 2009;106:8356–8361. doi: 10.1073/pnas.0900728106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Muthukumaraswamy SD, Evans CJ, Edden RA, Wise RG, Singh KD. Individual variability in the shape and amplitude of the BOLD-HRF correlates with endogenous GABAergic inhibition. Hum Brain Mapp. 2012;33:455–465. doi: 10.1002/hbm.21223. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Baslow MH, Hrabe J, Guilfoyle DN. Dynamic relationship between neurostimulation and N-acetylaspartate metabolism in the human visual cortex: evidence that NAA functions as a molecular water pump during visual stimulation. J Mol Neurosci. 2007;32:235–245. doi: 10.1007/s12031-007-0049-9. [DOI] [PubMed] [Google Scholar]
  27. Mangia S, Giove F, Dinuzzo M. Metabolic pathways and activity-dependent modulation of glutamate concentration in the human brain. Neurochem Res. 2012;37:2554–2561. doi: 10.1007/s11064-012-0848-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Zappe AC, Uludag K, Logothetis NK. Direct measurement of oxygen extraction with fMRI using 6% CO2 inhalation. Magn Reson Imaging. 2008;26:961–967. doi: 10.1016/j.mri.2008.02.005. [DOI] [PubMed] [Google Scholar]
  29. D'Esposito M, Deouell LY, Gazzaley A. Alterations in the BOLD fMRI signal with ageing and disease: a challenge for neuroimaging. Nat Rev Neurosci. 2003;4:863–872. doi: 10.1038/nrn1246. [DOI] [PubMed] [Google Scholar]
  30. Gustard S, Williams EJ, Hall LD, Pickard JD, Carpenter TA. Influence of baseline hematocrit on between-subject BOLD signal change using gradient echo and asymmetric spin echo EPI. Magn Reson Imaging. 2003;21:599–607. doi: 10.1016/s0730-725x(03)00083-3. [DOI] [PubMed] [Google Scholar]
  31. Lund TE, Norgaard MD, Rostrup E, Rowe JB, Paulson OB. Motion or activity: their role in intra- and inter-subject variation in fMRI. Neuroimage. 2005;26:960–964. doi: 10.1016/j.neuroimage.2005.02.021. [DOI] [PubMed] [Google Scholar]
  32. Liu X, Zhu XH, Chen W. Baseline BOLD correlation predicts individuals' stimulus-evoked BOLD responses. Neuroimage. 2011;54:2278–2286. doi: 10.1016/j.neuroimage.2010.10.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Lu H, Zhao C, Ge Y, Lewis-Amezcua K. Baseline blood oxygenation modulates response amplitude: physiologic basis for intersubject variations in functional MRI signals. Magn Reson Med. 2008;60:364–372. doi: 10.1002/mrm.21686. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Donahue MJ, Near J, Blicher JU, Jezzard P. Baseline GABA concentration and fMRI response. Neuroimage. 2010;53:392–398. doi: 10.1016/j.neuroimage.2010.07.017. [DOI] [PubMed] [Google Scholar]
  35. Patel A, de Graaf R, Mason G, Rothman D, Shulman R, Behar K. The contribution of GABA to glutamate/glutamine cycling and energy metabolism in the rat cortex in vivo. Proc Natl Acad Sci USA. 2005;102:5588–5593. doi: 10.1073/pnas.0501703102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Shulman RG, Hyder F, Rothman DL. Insights from neuroenergetics into the interpretation of functional neuroimaging: an alternative empirical model for studying the brain's support of behavior. J Cereb Blood Flow Metab. 2014;34:1721–1735. doi: 10.1038/jcbfm.2014.145. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Journal of Cerebral Blood Flow & Metabolism are provided here courtesy of SAGE Publications

RESOURCES